Web Appendix: Intergenerational Networks, Unemployment, and Persistent Inequality in South Africa

Literature › Works

Anon. (n.d.). Web Appendix: Intergenerational Networks, Unemployment, and Persistent Inequality in South Africa.

Type	other
Scope	Post-1994 South African economy
Method	econometric
Period studied	2002–2004
Regions	South Africa, South African provinces, urban enumeration areas
Themes	inequality poverty unemployment, labour markets wages, demography urbanisation, methods data sources, transition post apartheid
Citation key	anonnde

Argument

This web appendix supports a main paper, not included here, on how family and neighbourhood networks shape young South Africans' employment. It shows that sons' industries correlate with fathers' and daughters' with mothers', while cross-gender correlations are small and jointly insignificant. Occupational correlations are weaker and jointly insignificant for all four parent-child pairs. Attrition tests and bounding exercises suggest the network results are not an artefact of panel attrition. Further tests show that fathers' industry employment trends matter for network-found jobs but not self-found jobs, and that neighbourhood youth employment matters mainly for self-sought jobs. A final section explains how provincial industry employment series were built from household surveys and corrected for sampling error.

Evidence and method

A household panel (baseline 2002) of young black and coloured adults linked to parents' industries and occupations, with 687/649 (industry) and 829/819 (occupation) son/daughter observations. Provincial two-digit industry employment from the September Labour Force Surveys 2002-2004, which are nationally representative with 67,836-73,797 adults per round. The 1993 October Household Survey (91,494 adults) for comparisons between 1993 and 2003.

Findings

• Industry correlations are gendered, with same-gender parent-child links of roughly 7-12% in most industries. (p. 1)

  most industries being associated with an 7-12% increase in likelihood of employment in that industry for children if their parents work in that industry and happen to be the same gender.

• Occupational correlations between parents and children are jointly insignificant for all four parent-child pairs, unlike industries. (p. 1)

  parent’s occupations are not correlated with children’s across the sample in any of the 4 parent-child pairs

• The attrition analysis suggests attritors work less at baseline, making the none-working bound the most plausible case. (p. 2)

  Since attritors work less at baseline than non-attritors, we may think that column 2 represents the most likely case, where none of them are working.

• Neighbourhood youth employment strongly predicts young men's employment, though it likely proxies for many neighbourhood trends. (p. 3)

  This neighborhood youth employment appears very important for predicting young males’employment status

• Fathers' industry employment trends matter for network-found jobs and not for self-found jobs. (p. 3)

  the trends in fathers employment remain strong for network jobs and absent for self-reported jobs in the presence of the correlated networks controls

• The provincial industry employment series come from the 2002-2004 September Labour Force Surveys, constructed by the author. (p. 4)

  Provincial employment data are constructed by the author using two-digit industries from the September Labour Force Surveys in 2002-2004

• The survey-sampling-error correction barely affects the standard errors. (p. 4)

  In an unclustered equivalent of the baseline estimation, the largest standard error increases by 2.5 percent in magnitude

• Pre- and post-transition comparisons rely on the 1993 October Household Survey and the 2003 Labour Force Survey. (p. 5)

  Comparisons between 1993 and 2003 are conducted using the September 2003 Labour Force Survey and 1993 October Household Survey